When is a Sum not a Sum? - Cautions from the Riemann Rearrangment Theorem -

[Jfellers]

Intro: We were briefly introduced to the Riemann Rearrangement Theorem in my Calculus 2 class this week. The introduction left the class very confused because, at first glance, it seems to violate the basic nature of addition. How can you add the same series of numbers in two different arrangements and arrive at completely different answers? Shouldn't the sum be the same regardless of the order of addition? After all, any child could tell you that 3 + 1 = 4, just as 1 + 3 = 4. Curiously enough, when a series stretches off to infinity, strange things can happen that alter the sum and seemingly violate our fundamental understanding of addition.

Background: A sequence is just a list of numbers with some pattern connecting them. We usually build them with the answers to an equation we're working with. If f(x) = x+1, then we could write the answers as:

f(1) = 2, f(2) = 3, f(3) = 4.
Then the list:
a_n = [2, 3, 4, ...etc.]

would constitute the sequence of f(x) = x+1. Now, if we added the values of the sequence together, we could make it into a series.

S_n = 2 + 3 + 4 +... etc.

If we plotted the partial sums of the series on a graph, we would see each series behave in one of three ways: it will either grow to infinity (or negative infinity), it will absolutely converge and plateau at some value that it will never exceed (hit an asymptote), or it will conditionally converge. The conditionally convergent state is where the math abandons our human intuition.

For our series to be conditionally convergent, it must be convergent on its own. However, if we take the same sequence of numbers and modify it by taking the absolute value of each number, the resulting sum of the series will diverge to infinity. By contrast, if we made the same modification to a series that absolutely converges, the resulting series would still converge absolutely. That's why it's dubbed "Absolutely Convergent," while the trickier series are dubbed "Conditionally Convergent."

The conditionally convergent series behaves oddly because it is full of alternating positive and negative terms. The classic example used to show this pattern is the Alternating Harmonic Series:

S_n = 1-(1/2)+(1/3)-(1/4)+(1/5)-...etc.

The key takeaway is that conditional convergence depends on a delicate balance between infinitely many positive and infinitely many negative terms that effectively cancel each other out. The sum of the positive terms grows without bound, and the sum of the negative terms decreases without bound. The original order forces them to cancel in a specific pattern that initially causes convergence. If we rearrange that pattern, the cancellation can happen differently, and the final sum can change.

A useful analogy to imagine the difference in the counting procedure is your bank account. If you spend every Friday night before payday quietly attending to the business of your home, your paycheck hits first (+), and then your expenses follow (-) in a repeatable pattern:

+,-,+,-,...etc. = 0

But if you instead cut loose on Friday night and spend the money before the paycheck hits, the counting order changes and the balance of your account does as well:

-,-,+,- = - (Overdraft)

We can have radically different results depending on how we spend the weekend, or rather, the order in which we count the series.

Wisdom: The heart of the Riemann Rearrangement Theorem is that, for a conditionally convergent series, rearranging the terms can change the value of the series. The methodology determines the result; therefore, it can be skewed or altered to fit a narrative. The problem is not that ordinary addition has failed, but rather that an infinite series is not ordinary finite addition. It is a limit process, and limit processes can depend on the specific path used to approach infinity.

That lesson reaches beyond calculus. We see this every day. A car company might advertise an impressive horsepower or torque number that isn't false, but it's crafted to deceive you. They might reach that impressive performance on a manicured road, with a modified transmission and a lighter frame.

The theorem warns us to be wary of method-dependent conclusions. If a model relies on cherry-picked data or a highly precise methodology, chances are it's masking contrary evidence. Absolute convergence works in the theorem the same way truth works in our society. A truth worth trusting should not collapse the moment the terms are counted in a different order.

[StudentDriver]

Nice Post on the alternating Harmonic Series!

-if we let {bn}= 1, -(0.5), -0.25, 0.3333...

and {an}= 1, -0.5, 0.3333,-0.25....

I think there is a formal proof that the infinite summation of {a_n} equals to ln(2).

And the infinite summation of {b_n} = 0.5* ( 1-0.5+0.3333-0.25.....) = 0.5*ln(2)